The subject matter disclosed in the present specification relates to plural gateway devices which are installed in servers or in communication paths between terminals and the servers in a network system for communicating data between the servers and the terminals, such as the WWW (World Wide Web), a mail system and a data center.
Client terminals access servers coupled to LAN (Local Area Network) or WAN (Wide Area Network) through a lot of gateway devices such as switches, firewalls and gateways. Particularly, when clients such as mobile phone terminals access servers, an extremely large number of terminals access the servers through gateways. With the spread of terminals coupled to wired network and wireless network, high function of recent mobile phone terminals, high-speed operation of wireless communication network and a large capacity of contents such as animation and music, the amount of data exchanged between servers and clients in WWW or the like is increased. Further, the amount of communication is increased with reduction of data communication time due to increased capacity of communication band of wired network and wireless network and increased size of contents.
The amount of communication passing through gateway devices such as switches, firewalls and gateways in a data center system and a mobile phone operator system, or a telecommunication carrier system, is greatly expanded and is being increased. It is urgently required to reinforce the processing capability of gateways and servers due to such increased communication amount. As measures of reinforcing the capability, there are a method of improving hardware performance and a method of processing requests dispersedly.
Generally, the former is named scale-up and the latter is named scale-out. The measures using the scale-up have problems such as stop of service due to single defective point and stop of service upon update of hardware. A lot of mobile phone operators, or telecommunication carriers, and data center operation companies having large-scale system reinforce the capability of scale-out type capable of coping with increased communication amount without stopping service.
On the other hand, in order to attain high-speed responses to access requests from clients, servers have the data grid structure which utilizes high-speed access performance of a semiconductor memory to store data therein and makes responses instead of reading out data stored in a hard-disk to make responses. In the system constituted of plural servers utilizing the data grid structure, duplication and management of data between servers are performed by protocol having high scalability for securement of system reliability. In such a system, even if a request is issued to any server, the request is transferred to a server having data and accordingly target data can be accessed.
Such a system has a merit that the party that accesses data is not required to consciously consider where data is located. On the other hand, transfer of a request between servers in which data are provided occurs and there is a case where a response of system is delayed.
Moreover, as the system is more large-scale, data communication through plural gateway devices is also increased. Analysis is sometimes made until a high-rank layer according to kinds of gateway devices and transfer processing (latency as viewed from client terminals) of communication data possibly takes time.
Technique for solving the above problem is disclosed in JP-A-2010-183421. In this technique disclosed in this publication, when switches are provided at front and back stages in communication made through plural gateway devices, retrieval speed for retrieval table is improved when the switch at front stage makes routing by the switch at back stage, so that high-speed operation is realized.